Linux shell crash course¶

Note

This is a kickstart for the Linux shell, to teach the minimum amount needed for any scientific computing course. For more, see the linux shell course or the references below.

This is basic B-level: no prerequisites.

Watch this in video format

There is a companion video on YouTube, if you would also like that format (and a slightly longer one with more detail).

If you are reading this case, you probably need to do some sort of scientific computing involving the Linux shell, or command line interface. You may wonder why we are still using a command line today, but the answer is somewhat simple: once you are doing scientific computing, you eventually need to script and automate something. The shell is the only method that gives you the power to do anything you may want.

These days, you don’t need to know as much about the shell as you used to, but you do need to know a few important commands because the command line works when nothing else does - and you can’t do scripting without it.

What’s a shell?¶

It’s the old-fashioned looking thing where you type commands with a keyboard and get output to the screen. It seems boring, but the real power is that you can script (program) commands to run automatically - which is the point of scientific computing.

You type a command, which may include arguments. Output gets shown to the screen. Spaces separate commands and arguments. Example: cp -i file1.txt file2.txt. cp is the command, -i is an option, file1.txt and file2.txt are arguments.

Files are represented by filenames, like file.txt. Directories are separated by /, for example mydir/file.txt is file.txt inside of mydir.

Exercise: Start a shell. On Linux or Mac, the “terminal” application does this.

Editing and viewing files¶

nano is an editor which allows you to edit files directly from the shell. This is a simple console editor which always gets the job done. Use Control-x (control and x at the same time), then y when requested and enter, to save and exit.

less is a pager (file viewer) which lets you view files without editing them. (q to quit, / to search, n / N to research forward and backwards, < for beginning of file, > for end of file)

Listing and moving files¶

ls lists the current directory. ls -l shows more information, and ls -a shows hidden files. The options can be combined, ls -la or ls -l -a. This pattern of options is standard for most commands.

mv will move or rename files. For example, mv file.old file.new.

cp will make a copy of a file, with the exact same syntax as mv: cp file.old file.copy.

rm will remove a file: rm file.txt. To remove a directory, use rm -r. Note that rm does not have backups and does not ask for confirmation!

mkdir makes a directory: mkdir dirname.

Current directory¶

Unlike with a graphical file browser, there is a concept of current working directory: each shell is in a current directory. If you ls, it lists files in your current directory. If a program tries to open a file, it opens it relative to that directory.

cd dirname will change working directories for your current shell. Normally, you will cd to a working directory, and use relative paths from there. / alone refers to the root directory, the parent of all files and directories.

cd .. will change to the parent directory (dir containing this dir). By the same token, ../.. the parent of the parent, and so on.

Exercise: Change to some directory and then another. What do (cd -) and (cd with no arguments) do? Try each a few times in a row.

Online manuals for any command¶

man is an on-line manual, type man ls to get help on the ls command. The same works for almost any program. Some are easy to read, some are impossible. In general you look for what you need, not read everything. Use q to quit or / to search (n and N to search again forward and backwards).

--help or -h is a standard argument that prints a short help directly.

Exercise: briefly look at the manual pages and --help output for the commands we have learned thus far. How can you make rm ask before removing a file?

History and tab completion¶

Annoyed at typing so much? We’ve got two ways to make work faster.

First, each shell keeps its (shell) history. By pushing the up arrow key, you can access previous lines. Never type similar things twice, go up in history and find the previous line, modify it, then push enter to re-run.

Shells also have tab completion. Type the first few letters of any command or filename and push tab once or twice… it will either complete it or show you the options. This is so important that it’s used often, and many command arguments can also be completed.

Exercise: Play around with tab completion. Type pytho and push TAB. (erase that then start over) Then type p and push TAB twice. (erase that and start over) Then ls, space, and the first few letters of a filename, then push TAB.

Variables¶

There are two kinds of variables in shell: environment variables and shell variables. You don’t need to worry about the difference now. The $NAME or ${NAME} syntax is used to is used to access the value of a variable.

For example, the environment variable HOME holds your home directory, for me /home/rkdarst. The command echo prints whatever its arguments are, so echo $HOME prints my home directory. (Note that the variable is a property of the shell, not of the echo command - this is sometimes important).

To set a variable, use NAME=value. export NAME=value sets it as an environment variable which means that other processes you start (from this shell) can use it.

The $VARIABLE syntax is also often used for examples: in this case, it isn’t an environment variable, but just something you need to substitute yourself when running a command.

Quick reference¶

Cheatsheet

General notes

The command line has many small programs that when connected, allow you to do many things. Only a little bit of this is shown here.

Programs are generally silent if everything worked, and only print an error if something goes wrong.

ls [DIR]

List current directory (or DIR if given).

pwd

Print current directory.

cd DIR

Change directory. .. is parent directory, / is root, / is also chaining directories, e.g. dir1/dir2 or ../../

nano FILE

Edit a file (there are many other editors, but nano is common, nice, and simple).

mkdir DIR-NAME

Make a new directory.

cat FILE

Print entire contents of file to standard output (the terminal).

less FILE

Less is a “pager”, and lets you scroll through a file (up/down/pageup/pagedown). q to quit, / to search.

mv SOURCE DEST

Move (=rename) a file. mv SOURCE1 SOURCE2 DEST-DIRECTORY/ copies multiple files to a directory.

cp SOURCE DEST

Copy a file. The DEST-DIRECTORY/ syntax of mv works as well.

rm FILE ...

Remove a file. Note, from the command line there is no recovery, so always pause and check before running this command! The -i option will make it confirm before removing each file. Add -r to remove whole directories recursively.

head [FILE]

Print the first 10 (or N lines with -n N) of a file. Can take input from standard input instead of FILE. tail is similar but the end of the file.

grep PATTERN [FILE]

Print lines matching a pattern in a file, suitable as a primitive find feature, or quickly searching for output. Can also use standard input instead of FILE.

du [-ash] [DIR]

Print disk usage of a directory. Default is KiB, rounded up to block sizes (1 or 4 KiB), -h means “human readable” (MB, GB, etc), -s means “only of DIR, not all subdirectories also”. -a means “all files, not only directories”. A common pattern is du -h DIR | sort -h to print all directories and their sizes, sorted by size.

stat

Show detailed information on a file’s properties.

find [DIR]

find can do almost anything, but that means it’s really hard to use it well. Let’s be practical: with only a directory argument, it prints all files and directories recursively, which might be useful itself. Many of us do find DIR | grep NAME to grep for the name we want (even though this isn’t the “right way”, there are find options which do this same thing more efficiently).

| (pipe): COMMAND1 | COMMAND2

The output of COMMAND1 is sent to the input of COMMAND2. Useful for combining simple commands together into complex operations - a core part of the unix philosophy.

> (output redirection): COMMAND > FILE

Write standard output of COMMAND to FILE. < does the opposite, read input from a file.

type COMMAND or which COMMAND

Show exactly what will be run, for a given command (e.g. type python3).

man COMMAND-NAME

Browse on-line help for a command. q will exit, / will search (it uses less as its pager by default).

-h and --help

Common command line options to print help on a command. But, it has to be implemented by each command.